Universal MCP server enabling any LLM or AI agent to utilize expert skills from your local filesystem. Reduces context consumption through lazy loading. Works with Claude, Cline, and any MCP-compatible client.
git clone https://github.com/kdpa-llc/local-skills-mcp.gitLocal Skills MCP is a universal Model Context Protocol server that allows any LLM or AI agent—Claude, Cline, Continue.dev, or custom agents—to access expert skills stored in your local filesystem. Skills are auto-discovered from multiple directories including ~/.claude/skills/, project-specific folders, and custom paths, with full support for hot reloading when skills change. The server uses lazy loading to minimize context usage: skill names and descriptions load initially (~50 tokens each), while full skill content loads only when requested. This approach works across any MCP-compatible client while maintaining portability, letting you write skills once and use them across different AI systems without modification.
Install globally via npm (npm install -g local-skills-mcp) and add the MCP server to your client configuration file (mcp.json for Claude Code/Desktop, or via VS Code settings for Cline). Skills are auto-discovered from ~/.claude/skills/, ./.claude/skills/, ./skills, and custom directories. Create skills by asking your AI to use the included skill-creator skill, or manually create SKILL.md files with YAML frontmatter in any skill directory.
Create reusable expert instructions for specialized tasks across Claude, Cline, and other MCP clients
Organize project-specific skills in hidden or visible directories with automatic discovery
Reduce token consumption by lazy-loading skill content only when the AI agent requests it
Customize built-in skills by overriding them in your local skill directories
No install command available. Check the GitHub repository for manual installation instructions.
git clone https://github.com/kdpa-llc/local-skills-mcpCopy the install command above and run it in your terminal.
Launch Claude Code, Cursor, or your preferred AI coding agent.
Use the prompt template or examples below to test the skill.
Adapt the skill to your specific use case and workflow.
I'm using the Local Skills MCP server to help me with [SPECIFIC TASK]. I have a local skill file named [SKILL_FILE_NAME] that I want to use. The file is located at [FILE_PATH]. Please use this skill to [DESCRIBE TASK] and provide me with the results.
# Local Skill Execution Results ## Task: Analyze Customer Feedback ### Summary - **Positive Feedback**: 78% - **Negative Feedback**: 15% - **Neutral Feedback**: 7% ### Key Themes - **Product Quality**: Highly praised for durability and performance. - **Customer Service**: Mixed reviews; some customers reported slow response times. - **Pricing**: Generally considered fair, but some customers mentioned it was a bit high. ### Recommendations - **Improve Customer Service**: Implement a faster response system to address customer inquiries promptly. - **Maintain Product Quality**: Continue to focus on the high quality of the products to maintain customer satisfaction. - **Pricing Strategy**: Consider occasional discounts or promotions to attract more price-sensitive customers.
AI assistant built for thoughtful, nuanced conversation
Searchable AI prompt library for all major models
Cloud ETL platform for non-technical data integration
Get more done every day with Microsoft Teams – powered by AI
Customer feedback management made simple
Enterprise workflow automation and service management platform
Take a free 3-minute scan and get personalized AI skill recommendations.
Take free scan